Genomic evaluation of milk yield in a smallholder crossbred dairy production system in India

Abstract

Background India is the largest milk producer globally, with the largest proportion of cattle milk production coming from smallholder farms with an average herd size of less than two milking cows. These cows are mainly undefined multi-generation crosses between exotic dairy breeds and indigenous Indian cattle, with no performance or pedigree recording. Therefore, implementing genetic improvement based on genetic evaluation has not yet been possible. We present the first results from a large smallholder performance recording program in India, using single nucleotide polymorphism (SNP) genotypes to estimate genetic parameters for monthly test-day (TD) milk records and to obtain and validate genomic estimated breeding values (GEBV). Results The average TD milk yield under the high, medium, and low production environments were 9.64, 6.88, and 4.61 kg, respectively. In the high production environment, the usual profile of a lactation curve was evident, whereas it was less evident in low and medium production environments. There was a clear trend of an increasing milk yield with an increasing Holstein Friesian (HF) proportion in the high production environment, but no increase above intermediate grades in the medium and low production environments. Trends for Jersey were small but yield estimates had a higher standard error than HF. Heritability estimates for TD yield across the lactation ranged from 0.193 to 0.250, with an average of 0.230. The additive genetic correlations between TD yield at different times in lactation were high, ranging from 0.846 to 0.998. The accuracy of phenotypic validation of GEBV from the method that is believed to be the least biased was 0.420, which was very similar to the accuracy obtained from the average prediction error variance of the GEBV. Conclusions The results indicate strong potential for genomic selection to improve milk production of smallholder crossbred cows in India. The performance of cows with different breed compositions can be determined in different Indian environments, which makes it possible to provide better advice to smallholder farmers on optimum breed composition for their environment.

Detection of genomic regions that differentiate Bos indicus from Bos taurus ancestral breeds for milk yield in Indian crossbred cows

“…In the regions sampled, Holstein-Friesian (HF) and Jersey were the only B. taurus breeds known to have been used for crossbreeding. After data quality control (QC) ( Al Kalaldeh et al, 2021 ), 11,510 crossbred cows with 223,379 TD yield in 5,305 herds remained. The number of crossbred cows per herd ranged from 1 to 44, with an average of 2.17.…”

“…The number of crossbred cows per herd ranged from 1 to 44, with an average of 2.17. To obtain adjusted TD milk yield, TD records were corrected for fixed effects using a mixed linear model as described in ( Al Kalaldeh et al, 2021 ). Fixed effects included parity, CDC (Cattle Development Centre), year-month (2016-09–2020-07), interaction of CDC and year-month, milk curve for each parity modelled by the third order Legendre polynomial (LP), milk curve for each CDC modelled by the third order LP.…”

“…After being subject to QC as described by (2021), animals genotyped with the 50K chip were imputed to high density (HD) using reference samples that had real HD genotypes. The reference population for imputation included 684 pure Jerseys, and 968 pure Holsteins from the Bovine HapMap (Consortium et al, 2009), 666 pure Indian B. indicus indigenous animals from a separate BAIF project (Strucken et al, 2021), and the 643 crossbred cows from the BAIF EGP. The HapMap and the Indian indigenous samples were genotyped with the Illumina 777k assay and genotypes were received post-QC.…”

“…We were unable to differentiate different sources of Desi using SNP marker data because of the remarkably low SNP variation observed between B. indicus breeds. Approximately 1% of all SNP variation within B. indicus occurs between breeds, compared to the nearly 40% of variation within B. taurus being between B. taurus breeds (Strucken et al, 2021). The GWAA for QTL segregating within B. taurus in the two-ancestor model, or within HF and Jersey in the three-ancestor model, had a power equivalent to undertaking a GWAA in approximately 3,200, 2,450 and 750 purebred cows, respectively.…”

“…Furthermore, no regions differentiating HF from Jersey reached the FDR30% compared to a large number of regions differentiating indigenous from exotic that reached the FDR30% due to the relatively lower power to detect loci differentiating HF from Jersey compared to between indigenous and exotic breeds. This is due to much lower genetic variation within B. taurus breeds compared to between B. indicus and B. taurus breeds (Strucken et al, 2021). Confirmation awaits more data from these Indian crossbred populations, which will be collected in the next several years, or analysis of other populations.…”